Differential D2550 Diff 9264 omnitools/src/Strings/Multibyte/OmniString.php

Changeset View

Standalone View

omnitools/src/Strings/Multibyte/OmniString.php

<?php		<?php
declare(strict_types=1);		declare(strict_types=1);

namespace Keruald\OmniTools\Strings\Multibyte;		namespace Keruald\OmniTools\Strings\Multibyte;

use Keruald\OmniTools\Collections\Vector;		use Keruald\OmniTools\Collections\Vector;

		/**
		* Represents a multibyte string and perform operations with the grapheme
		* library for UTF-8 encoding, and mbstring for other encodings.
		*/
class OmniString {		class OmniString {

use WithEncoding;		use WithEncoding;

///		///
/// Private members		/// Private members
///		///

/**		private string $value;
* @var string
*/
private $value;

///		///
/// Constructor		/// Constructor
///		///

public function __construct (string $value = '', string $encoding = '') {		public function __construct (string $value = '', string $encoding = '') {
$this->value = $value;		$this->value = $value;
$this->setEncoding($encoding ?: "UTF-8");		$this->setEncoding($encoding ?: "UTF-8");
Show All 28 Lines	class OmniString {
public function startsWith (string $start) : bool {		public function startsWith (string $start) : bool {
return str_starts_with($this->value, $start);		return str_starts_with($this->value, $start);
}		}

public function endsWith (string $end) : bool {		public function endsWith (string $end) : bool {
return str_ends_with($this->value, $end);		return str_ends_with($this->value, $end);
}		}

		/**
		* @deprecated Use more specific method to express your intent:
		* countBytes, countCodePoints or countGraphemes
		*/
public function len () : int {		public function len () : int {
		return $this->countGraphemes();
		}

		public function countBytes () : int {
		return strlen($this->value);
		}

		public function countCodePoints () : int {
return mb_strlen($this->value, $this->encoding);		return mb_strlen($this->value, $this->encoding);
}		}

public function getChars () : array {		public function countGraphemes () : int {
		return match ($this->encoding) {
		"UTF-8" => grapheme_strlen($this->value),
		default => $this->countCodepoints(),
		};
		}

		public function getBytes() : array {
		return str_split($this->value, 1);
		}

		public function getCodePoints () : array {
		return mb_str_split($this->value, 1, $this->encoding);
		}

		public function getGraphemes () : array {
		if ($this->encoding !== "UTF-8") {
		return $this->getCodePoints();
		}

$chars = [];		$chars = [];

$len = $this->len();		$len = grapheme_strlen($this->value);
for ($i = 0 ; $i < $len ; $i++) {		for ($i = 0 ; $i < $len ; $i++) {
$chars[] = mb_substr($this->value, $i, 1, $this->encoding);		$chars[] = grapheme_substr($this->value, $i, 1);
}		}

return $chars;		return $chars;
}		}

		/**
		* @deprecated Use more specific method to express your intent:
		* getBytes, getCodePoints or getGraphemes
		*/
		public function getChars () : array {
		return $this->getGraphemes();
		}

public function getBigrams () : array {		public function getBigrams () : array {
		return match ($this->encoding) {
		"UTF-8" => $this->getBigramsFromGraphemes(),
		default => $this->getBigramsFromCodePoints(),
		};
		}

		private function getBigramsFromGraphemes() : array {
		$bigrams = [];

		$len = grapheme_strlen($this->value);
		for ($i = 0 ; $i < $len - 1 ; $i++) {
		$bigrams[] = grapheme_substr($this->value, $i, 2);
		}

		return $bigrams;
		}

		private function getBigramsFromCodePoints() : array {
$bigrams = [];		$bigrams = [];

$len = $this->len();		$len = mb_strlen($this->value, $this->encoding);
for ($i = 0 ; $i < $len - 1 ; $i++) {		for ($i = 0 ; $i < $len - 1 ; $i++) {
$bigrams[] = mb_substr($this->value, $i, 2, $this->encoding);		$bigrams[] = mb_substr($this->value, $i, 2, $this->encoding);
}		}

return $bigrams;		return $bigrams;
}		}

///		///
Show All 35 Lines